Subject: Re: WARNING: SPL NOT LOWERED ON SYSCALL EXIT
To: None <current-users@netbsd.org>
From: Christos Zoulas <christos@zoulas.com>
List: current-users
Date: 02/15/2000 20:02:03
In article <x6r9eexnov.fsf@reddwarf.rightnowtech.com>,
Chris Jones <chris@cjones.org> wrote:
>Judging by the all-caps, this is probably a kernel message I should be
>concerned about...  :)
>
>I'm running i386 on a laptop, -current from last week sometime.  Using
>SOFTDEP on all fs'es.  Immediately following the message above, I get
>dropped into ddb.  Thus far, I've just been hitting "c," and getting
>on with life.  This is probably a bad idea, but I'm at a loss as to
>what SPL even stands for...  :)
>
>This seems to be caused by excessive disk activity, like deleting
>large directory trees (or registering packages).
>
>Any advice?  I'm quite willing to debug this using ddb or whatever...
>I'll post a stack trace RSN.
>
>Chris
>
>P.S.:  I'm about three months behind on reading current-users, so
>       please forgive me if this has already been reported.
>

Yeah, this is bad and it indicates that there is a bug in the kernel
where some syscall did int x = splfoo(); and did not call splx(x);
before it returned. I think that the stack trace might show you
the offending syscall(). in ddb type 't'.

oh, and spl stands for set processor level, meaning set the level
of interrupts you want to process, masking the rest.